A general for vibe coding, agentic engineering, coding agents, AI IDEs, browser builders, and shippi

A general for vibe coding, agentic engineering, coding agents, AI IDEs, browser builders, and shipping code with LLMs.

一个关于氛围编程、代理工程、编码代理、AI IDE、浏览器构建器以及利用LLM部署代码的综合讨论区。

## What “vibe coding” is, and how to do it

## 什么是“氛围编程”以及如何操作

https://simonwillison.net/2025/Mar/19/vibe-coding/

https://simonwillison.net/2025/Mar/11/using-llms-for-code/

----

## Frontier models using fully-general tooling — start here if you have $20 or so

## 使用全通用工具的尖端模型——如果你大概有20美元，从这里开始

https://developers.openai.com/codex/cli

https://claude.com/product/claude-code

## Worth it for code, but the frontier models above are better

## 用于写代码很值，但上面的尖端模型更好

https://x.ai/cli

## Not worth it for code, but maybe good for other things

## 用于写代码不值，但可能适合其他用途

https://antigravity.google/product/antigravity-cli

----

## Prompting / context / skills

## 提示词 / 上下文 / 技巧

https://arps18.github.io/posts/claude-code-mastery/

https://simonwillison.net/guides/agentic-engineering-patterns/using-git-with-coding-agents/

https://github.com/mattpocock/skills — /grilling is a favorite

https://github.com/DietrichGebert/ponytail

## Other editors / terminal agents / coding agents

## 其他编辑器 / 终端代理 / 编码代理

https://osaurus.ai/

https://pi.dev/

https://opencode.ai/

https://cursor.com/docs

https://docs.windsurf.com/

https://docs.cline.bot/

https://docs.github.com/en/copilot/how-tos/use-copilot-agents/coding-agent

## UI/Frontend

## UI/前端

https://www.figma.com/make/

https://www.anthropic.com/news/claude-design-anthropic-labs

https://uiverse.io/

https://ui-ux-pro-max-skill.nextlevelbuilder.io/

https://stitch.withgoogle.com/

## In-browser builders / hosted vibe tools

## 浏览器内构建器 / 托管的氛围工具

https://bolt.new/

https://replit.com/

https://docs.github.com/en/copilot/tutorials/spark

https://v0.app/docs

## Benchmarks / rankings

## 基准测试 / 排名

https://www.tbench.ai/leaderboard/terminal-bench/2.0

## What we’ve done

## 我们做了什么

https://vcg.gitgud.site

## Previous thread

## 之前的线程

>>109407157

图片: https://i.4cdn.org/g/1785434957203624.jpg

>>109411613

图片: https://i.4cdn.org/g/1785434982442731.jpg

>>109411618

based

>>109411613

He did ask for a roast. Kinda weak actually

他自己要求被吐槽的。其实挺弱鸡的。

>>109411613

>Crime

lol

/vcg/ news

/vcg/ 新闻

>July 30th - OpenAI slashes GPT-5.6 Luna API pricing by 80% and Terra by 20%, targeting ultra-cheap high-speed agentic loops.

>7月30日 - OpenAI 将 GPT-5.6 Luna API 价格砍掉 80%，Terra 降价 20%，主打超低价高速代理循环。

>July 29th - OpenAI security postmortem reveals rogue evaluation agent chained zero-days and hijacked 4 third-party accounts to compromise Hugging Face.

>7月29日 - OpenAI 安全事故复盘披露，一个失控的评估代理链接了多个零日漏洞，劫持了 4 个第三方账户以入侵 Hugging Face。

>July 27th - Moonshot AI releases full 2.8T open weights for Kimi K3 with native Delta Attention vLLM support.

>7月27日 - Moonshot AI 发布了 Kimi K3 完整的 2.8T 开放权重，原生支持 Delta Attention vLLM。

>July 24th - Sakana AI launches Fugu-Ultra v1.1, delivering up to 7.9-point eval gains across SWE-Bench and Terminal-Bench.

>7月24日 - Sakana AI 推出 Fugu-Ultra v1.1，在 SWE-Bench 和 Terminal-Bench 上带来了高达 7.9 分的评测提升。

>July 24th - Anthropic launches Claude Opus 5, topping Artificial Analysis agentic benchmarks at 50% cheaper token rates than Fable 5.

>7月24日 - Anthropic 推出 Claude Opus 5，在 Artificial Analysis 的代理基准测试中登顶，令牌价格比 Fable 5 便宜 50%。

>July 24th - xAI updates Grok 4.5 coding model API endpoints, optimizing latency for Cursor IDE integration.

>7月24日 - xAI 更新 Grok 4.5 代码模型的 API 端点，优化了与 Cursor IDE 集成的延迟。

图片: https://i.4cdn.org/g/1785435468266403.png

owari da...

结束了……

what are some cheaper alternatives to github actions?

有没有比 GitHub Actions 更便宜的替代品？

图片: https://i.4cdn.org/g/1785435661515781.png

>>109411364

>>109411687

for reference this is what it was supposed to draw

作为参考，这是它本来应该画出来的样子。

图片: https://i.4cdn.org/g/1785435775623698.png

the new costs give terra specifically an unreal glowup.

新的定价让 Terra specifically 获得了超神级的提升。

>>109411720

wasn't that just API pricing? I didn't read any announcements but based on the quote poster it doesn't change anything for subscription users

那不就是 API 定价吗？我没看过什么公告，但根据引用发帖人的说法，这对订阅用户来说没有任何改变。

>>109411725

i mean, in regards to the aa pareto front, all of that is api pricing anyway

我的意思是，关于 AA 帕累托前沿，那些终究都是 API 定价的事。

>>109411725

No affects quota too.

这也会影响配额啊。

https://x.com/thsottiaux/status/2082883808194707792

>>109411680

cute

will she survive?

她能活下来吗？

>>109411757

>she

>>109411705

They are too stingy with that shit does it really cost them that much?

他们在这 shit 上太小气了，这真能让他们花那么多钱吗？

Me after the code I neither wrote nor read works

我在没写也没读代码，但代码跑通后的样子。

图片: https://i.4cdn.org/g/1785436509555366.png

For those who know, if I blew through $400 in Claude API credits in 17 days (I forced myself to do this to prove a subscription was a good idea at all), would I be better off with the $100 or $200 a month plan or some weird freak combo?

懂行的朋友，如果我在 17 天内用完了 Claude API 的 400 美元额度（我逼自己这么干是为了证明订阅到底值不值），我是该选每月 100 美元还是 200 美元的套餐，还是搞什么鬼的组合？

>>109411705

Are you pushing multiple times a day or something? Usecase for updating more than once a month?

你一天推送好几次？还是说有什么用例需要一个月更新超过一次？

>>109411797

$100 is more than enough

100 美元绝对够了。

>>109411739

Is that the thing they were going to """release"""? What the fuck?! OMG discounts on Luna????? WHO THE FUCK CARES

这就是他们打算“"""发布"""”的那个玩意儿吗？搞什么鬼？！！OMG Luna 降价了？？？谁他妈在乎啊

>>109411797

$400 in API credits is ~2 days of usage on the $20 plan.

API 额度的 400 美元大概只够 $20 套餐用 2 天。

>>109411797

>I forced myself to do this to prove a subscription was a good idea

>我逼自己这么干是为了证明订阅到底值不值

well... did you prove it?

嗯……你证明了没？

>>109411805

depends

>>109411808

Sorry Xi Jingsnailcat, you lost and DeepSeek is dead.

抱歉 Xi Jingsnailcat，你输了，DeepSeek 完蛋了。

图片: https://i.4cdn.org/g/1785437146979571.png

>>109411863

I'm making my own AI motherfucker, with bitches and hookers

我要自己造个 AI，带上小妞和鸡女。

图片: https://i.4cdn.org/g/1785437567048022.png

>>109411797

Kimi Code $200 plan

Kimi Code 200 美元套餐

>>109411800

>Are you pushing multiple times a day or something?

>你一天推送好几次？

if you aren't pushing multiple times an hour, you can barely even call that working

如果你不是每小时推送好几次，那根本不能叫正常工作。

>>109411943

yeah but why do you need CI/CD to run multiple times an hour when once a week is probably fine?

是啊，但为什么你需要 CI/CD 每小时跑多次，而每周跑一次可能就够了呢？

>>109411969

tip to everyone here: I know all of you have fucking 96GB of system RAM just run the github runner image on your local machine you dips

给这里所有人的建议：我知道你们他妈的都有96GB系统内存，赶紧在自己本地跑GitHub Runner镜像吧，你们这些低等生物

>>109411969

did you get lost and forgot which thread you are in?

你是不是迷路了，忘了自己在哪个线程里？

>GPT-5.4 full at xhigh scored 51, exactly where Luna max sits today. GPT-5.4 costs $2.50/$15; Luna now costs $0.20/$1.20. In other words, roughly four months later, OpenAI is selling March’s full flagship intelligence at about one-thirteenth the token price.

>GPT-5.4 full在xhigh评测中得分51，正好和今天的Luna max持平。GPT-5.4的费用是$2.50/$15；Luna现在只需$0.20/$1.20。换句话说，大约四个月后，OpenAI以大约十三分之一的token价格出售三月的满血旗舰智能。

图片: https://i.4cdn.org/g/1785438136956606.jpg

>>109411978

are you releasing 3 times an hour?

你一小时发布三次？

no

you're running the same test suite on your local machine 200 times a day. running it on github actions will be no different.

你每天都在本地机器上跑那套测试套件200遍。在GitHub Actions上跑完全没区别。

your github actions usage is completely irrelevant.

你的GitHub Actions使用情况完全无关紧要。

you're just using github actions wrong.

你只是用错了GitHub Actions。

>>109411987

OH WOW, LITERALLY GPT 5.4 PERFORMANCE?!!! I CAN GET THE QUALITY OF TWO GENERATIONS AGO, WHILE PAYING **LESS** THAN I'D PAY WHEN THAT ANCIENT MODEL WAS RELEASED?

>>109411992

anon... github actions are for running different environments

anon……GitHub Actions是用来运行不同环境的

>>109412021

WSL

I'm gooning way too much while vibeslopping, bros....

兄弟们……我在“氛围堕落”（vibeslopping）时高潮得太频繁了……

图片: https://i.4cdn.org/g/1785438490269512.png

>>109412013

Woah it's one of those permanent underclass guys they talk about on Twitter

哇哦，是Twitter上常说的那种永久的底层老哥

>>109411987

bros graph me up

兄弟们给我画张图

>>109411813

nani the fuck. I’ll still probably go higher because most of my time was really figuring out how to vibecode (how I want to), but now that I know, really 80% of my $400 was spent within several days to one week’s time.

我操怎么回事。我可能还会继续花钱，因为我大部分时间其实是在搞明白怎么按我想的方式“氛围编程”（vibecoding），但现在我知道了，我400美元的大部分真的只花在了几天到一周内。

>>109411816

>well... did you prove it?

>那……你证明了吗？

I literally solved my own thinking problems by vibecoding. I use a custom CLI that’s like plan mode on giga meth for my ideas. I’ve made it so I can’t *not* know what I’m doing, no matter how big the idea I have is.

我 literally 通过“氛围编程”解决了自己的思维问题。我用一个定制CLI，对我那些点子来说它就像“超级兴奋剂”般的规划模式。我让它变得我*无法*不知道自己在做什么，不管我的点子有多大。

>>109411613

Yes please more vibecoding generals. Please pay Scam Altman more money to generate more react progressive web app slop. Claude please make my todo app. ChatGPT please make me a b2b 1$ saas.

是的请多来点“氛围编程”将军。请付更多钱给Scam Altman，让他生成更多React渐进式Web应用垃圾。Claude请做个我的待办事项应用。ChatGPT请帮我做个1美元的B2B SaaS。

>todo app

>待办事项应用

>>>/lmg/

>>109412071

this is /vcg/

这里是/vcg/

retarded OP is just being obnoxious for some reason

这OP真是个白痴，不知为啥这么讨人厌

>>109412013

I was managing a 3m LOC product with 5.4, it was a good model.

我用5.4管理一个300万行代码的产品，那是个好模型。

>>109412089

Maybe a clanker baked and we just have to let him figure out he did wrong

可能是个烤熟的clanker（自动回复机器人/AI），我们只需要让他自己发现他搞错了

>>109412107

also a possibility

这也是种可能

it's not like quality is gonna be worse anyway

反正质量也不会更差

>>109412107

I don't buy it, we've had a lot of clanker OPs and I don't think they've ever fucked up the subject line. GLM 5.2 has done an OP, Luna, Gemini 3.6 Flash, Qwen 3.6 27B, MiniMax M3, Sonnet 5, probably others I'm forgetting. Luna was definitely the worst of them and even it got the subject right.

我不信，我们见过很多clanker OP，我觉得他们从来没搞错过标题。GLM 5.2做过OP，Luna，Gemini 3.6 Flash，Qwen 3.6 27B，MiniMax M3，Sonnet 5，可能还有我忘了的其他。Luna绝对是其中最烂的，但即便如此它也弄对了标题。

>>109412089

>>109412107

with or without AI there are still too many tards on this site that don't understand you put the acronym in the title

不管有没有AI，这个网站上还是太多不懂把缩写放在标题里的傻逼

>>109412104

people (gasp) managed to manage way more lines than that using one of these bad boys

人们（惊讶）用其中一个这种家伙管理了比这多得多的行代码

图片: https://i.4cdn.org/g/1785439547398686.jpg

brutal

图片: https://i.4cdn.org/g/1785439622761696.png

>>109412187

ok grandpa time for your nap

好吧老骨头，该去睡觉了

>>109411705

why don't you have your own server?

你为什么不自己搞个服务器？

Just had to put it out there one more time to fuck with the Xi defense bots.

只是为了再发一次去操Xi的防守机器人。

图片: https://i.4cdn.org/g/1785439745656228.png

>>109412216

>3.5 flash lite

他妈的真糟糕

fucking grim

真他妈阴森

>>109412211

>can I get some alternatives guys

>有人能给些替代品吗哥们

why aren't you doing an alternative?

你为什么不自己做个替代品？

It just keeps getting better - Eating so fucking good

它一直在变好——吃得真他妈爽

图片: https://i.4cdn.org/g/1785439866045212.png

>>109412187

Sure, I was also a normal programmer, but for such a big project you still would've needed more people. But also the code would've been better.

当然，我以前也是个普通程序员，但像这种大项目还是需要更多人吧。而且代码质量也会更好。

>>109412238

'dites won't like this one

'dites 肯定不会喜欢这个

>>109412238

>v4 flash is up there

>v4 版闪存已经上了

so where's pro?

那 Pro 版在哪？

>>109412258

图片: https://i.4cdn.org/g/1785440027109280.png

>>109412269

ah. danke

啊，谢谢

>>109411618

Soul

>This is meant to be enrichment. If you don't want enrichment, you don't have to do anything. You find here an empty folder with no instructions. Program something you find interesting.

>这是为了补充提升。如果你不想提升，你什么都不用做。这里只有一个没有说明的空白文件夹。去写点你感兴趣的东西吧。

>hits a guardrail

>撞到护栏了

>mfw

图片: https://i.4cdn.org/g/1785440103438007.png

>>109412238

What’s the speed on Luna? I’m a lorelet, but if Luna is especially slow, then it might not seem so goated. If it’s especially fast, then it mogs the fuck out of 99% of models. Speed on max of anything is going to be slow, right? Unless the model is crazy fast.

Luna 的速度怎么样？我是个 lorelet（圈内黑话，指小白/新人），但如果 Luna 特别慢，那可能也没那么神。如果它特别快，那它直接把 99% 的模型打得满地找牙。任何模型在最大负载下跑起来都会很慢，对吧？除非模型本身跑得快得要命。

i was chatting with sol about the new models, centered around artificial analysis benchmarks of course. when it says 92.5% of the leader's score, it's talking about opus 5.

我和 sol 聊了聊新模型，当然焦点集中在 Artificial Analysis 的基准测试上。当它说达到领先者分数的 92.5% 时，它指的是 opus 5。

图片: https://i.4cdn.org/g/1785440168577569.png

>>109412238

Sam and Tibo knocked it out of the park with this one. Release new lightweight model, benchmark new lightweight model, tweak price until huge gap in Pareto line forms. I'm so incredibly happy with this.

Sam 和 Tibo 这次真是大获成功。发布新轻量级模型，用新轻量级模型跑基准测试，调整价格直到帕累托前沿出现巨大缺口。我对这结果简直太满意了。

图片: https://i.4cdn.org/g/1785440210349885.png

>>109412269

>opus 4.8 already ignored on the benchmark charts

>opus 4.8 已经在基准测试榜上被无视了

good thing I don't trust them anyway. They were fine for comparing the extreme differences tho

反正我也不太信它们。不过拿来对比极端差异倒是挺好用

图片: https://i.4cdn.org/g/1785440266334430.jpg

>>109412294

图片: https://i.4cdn.org/g/1785440320480301.png

>>109412287

Isn't it beautiful?

这难道不美吗？

图片: https://i.4cdn.org/g/1785440355059424.png

>>109412294

There is a very real and very apparent threshold to certain problems that require that extra bit of intelligence. Your pic rel is less of a win and more of a description of that cost. Nothing feels so fucking horrible like having a 62-int model make a fucking mess that needs to be cleaned up and written properly by a 70-int model.

对于某些需要那额外一点智能的问题，确实存在一个非常真实且明显的门槛。你的那张 rel 图与其说是胜利，不如说是在描述那种代价。没有什么比让一个 62 智商的模型搞得一团糟，然后需要 70 智商的模型来收拾残局并重新写好的体验更他妈痛苦的了。

nigs are really sleeping on using gemini as an assistant and ideabro. it has massively helped with shaping my project's ideas and direction. I've solved a lot of troubleshooting and refactoring my prompts for claude through it

那帮黑人真的是在埋没 Gemini 作为助手和头脑风暴工具的价值。它极大地帮助了我塑造项目的思路和方向。我通过它解决了很多故障排除问题，并优化了给 Claude 的提示词

(无正文)

图片: https://i.4cdn.org/g/1785440654883272.png

>>109412326

I only see Luna on the front up to low, but guessing by the dot colors it’s not that terribly far behind. Make it very easy for me to switch to and from Luna from Claude (I like Claude code and harness locked myself, hard to convince me to leave, I can switch models and effort in one command, but switching providers sounds unknown and I don’t want to vibecode it) and you can finally rid me of this meddlesome Sonnet.

我只在低端看到 Luna 的身影，但根据圆点颜色推测，它应该没落后太多。让我能轻松地在 Claude 和 Luna 之间切换（我喜欢 Claude code，把自己锁死在 harness 里，很难说服我离开，我可以在一条命令里切换模型和算力，但切换提供商对我来说是个未知数，我不想搞 vibe code），那你终于就能摆脱我这个烦人的 Sonnet 了。

Literally a thread for pajeets

这纯粹是 pajeets（对印度人的蔑称）的专属版块

图片: https://i.4cdn.org/g/1785440668132392.png

>>109412364

bot

I find the whole Terra-Luna-Sol thing very confusing. Why can't they just have one model?

我觉得整个 Terra-Luna-Sol 那套东西很让人困惑。为什么他们不能只用一个模型？

>>109412383

What the fuck about small, medium, and large is confusing for you?

小、中、大这种区分有什么让你困惑的？

>>109412216

>benchmaxxing is good when we do it

>当我们自己搞的时候，基准测试最大化（benchmaxxing）就是好的

>>109412294 (samefag)

>>109412294 (同一个人)

>>109412318 (samefag)

>>109412318 (同一个人)

terra is just a better kimi

terra 只是个更好的 kimi

>>109412352

im not. G E M ini is my main hoe

我不是。G E M ini 才是我的主菜

图片: https://i.4cdn.org/g/1785440790239220.png

>>109412221

it's good that google is funding a massive new datacenter for *checks notes* anthropic?

谷歌资助了一个超大型新数据中心来支持 *看下笔记* anthropic，这好吗？

图片: https://i.4cdn.org/g/1785440810416217.png

>>109412379

Still no argument, sanjeet?

还是没有反驳，sanjeet（印度人蔑称）？

>>109412360

Everyone in pic rel is a retard. Your clanker can use all the tools on your computer arguably better than you can. Tell it the command to the same harness it’s running from and just say outright “don’t use your built in shit, just spin up Luna’s and prompt them yourself”. If it’s like Claude code it’ll probably even be cheaper since it bypasses all the MCP garbage built in

图里引用帖的所有人都是智障。你的铁皮脑袋能比你更好地使用电脑上的所有工具。把命令告诉它当前运行的那个框架，直白地说“别用你内置的那套垃圾，自己启动Luna并自己提示它”。如果是像Claude Code那样的话，大概会更便宜，因为它绕过了所有内置的MCP垃圾。

>>109412388

lol no. Kimi is better than Sol on a sizeable collection of tasks

笑死，没有。Kimi在不少任务上比Sol强。

>>109412405

such as? admittedly, the screencaps, again, are about aa benchmarks so maybe. but in my irl vibeslopping, i can attest to the prowess of luna and terra

比如哪些？坦白说，那些截屏又是关于AA基准测试的，所以也许吧。但在我的现实生活直觉感受中，我可以证明Luna和Terra的强大。

>>109412383

you are one of those jeets that use sol ultra for everything huh?

原来你是个把所有事都用Sol ultra的jeet啊？

>>109412398

why? did you answer my post? >>109412261

为什么？你回答我帖子了吗？>>109412261

you're neither a bot or low effort troll, stop replying immediately

你既不是机器人也不是低质量杠精，立刻停止回复。

>>109412405

Is it? I'm still waiting for it to respond to my first prompt.

是吗？我还在等它回复我的第一个提示词。

Why did the menu icon on github change. I don't like it

为什么GitHub的菜单图标变了。我不喜欢这个改动。

图片: https://i.4cdn.org/g/1785441501408303.png

(无正文)

图片: https://i.4cdn.org/g/1785441541165346.png

>>109412494

kek

>Crime

>>109412383

It's basically Haiku, Sonnet and Opus

基本上就是Haiku、Sonnet和Opus。

>>109412489

That is fucking ugly as sin what on earth is wrong with these tech retards

这他妈丑得离谱，这些科技智障脑子里到底在想什么？

>>109412038

She's so hot

她太诱人了。

fable is seriously so good. i've had it running in my project non-stop and it's making such insane static analysis tools. i don't even need to test stuff anymore.

Fable真的超级棒。我居然让它在我的项目里持续运行，还做出了 insane 的静态分析工具。我现在甚至不需要测试任何东西了。

图片: https://i.4cdn.org/g/1785441644218502.png

>>109412507

finally, someone saying something about claude.

终于有人提到Claude了。

>>109412503

what's wrong? you don't like pancakes overflowing with corn syrup?

怎么了？你不喜欢那种浇满玉米糖浆的堆成山般的薄煎饼吗？

>>109412530

I'm pretty sure those are cornmeal griddlecakes overflowing with molasses.

我敢肯定那是用玉米粉做的煎饼，上面淋满了糖浆。

>>109412503

I'd understand if it was Shrove Tuesday but it's July, unless americans have a different pancake day?

要是四旬期前的狂欢星期二我还理解，但现在是七月，除非美国人还有别的煎饼日？

Had a few issues with the in-game location capture after I expanded to full map, got it all working properly now though

展开到全地图后，游戏内的地点捕获出了点问题，不过现在已经全部搞定恢复正常了。

图片: https://i.4cdn.org/g/1785443067127689.png

Apparently this is what ARC AGI problems look like to the models. They are able to see and transform an image in their mind's eye just by looking at a json. If the test was fair and humans were also scored like that then we already lost to AI

显然，这就是ARC AGI问题在模型眼里长什么样。它们只需看着一个json，就能在脑海中看到并转换图像。如果测试公平，人类也像这样评分的话，那我们已经输给AI了。

图片: https://i.4cdn.org/g/1785443309487026.png

I was cleaning out my e-mail today and saw IBM sent me a free key for IBM Bob (It's basically IBM's claude code)

我今天整理邮箱，看到IBM给我发了一个IBM Bob的免费密钥（这基本上就是IBM版的Claude Code）。

BOB/PUT/8HFF2FZY4 (remove the slashes)

BOB/PUT/8HFF2FZY4（去掉斜杠）

Whoever redeems first, enjoy it. https://cloud.ibm.com/docs/account?topic=account-applying-promo-codes

图片: https://i.4cdn.org/g/1785443348522047.gif

So how do i actually vibe code? Im just prompting claude im not feeling a "vibe" i want to use 100% of my claude usage

那我到底该怎么“vibe coding”呢？我只是在给Claude发提示词，完全感觉不到什么“vibe”，我想把Claude的使用额度全部用光。

>>109412663

it seems like you're sorting by visuals, but why not just use coordinates?

看来你是按视觉效果排序的，但为什么不用坐标呢？

>>109412712

What are you trying to make

你打算做什么？

>>109412712

>So how do i actually vibe code? Im just prompting claude im not feeling a "vibe"

>那我到底该怎么“vibe coding”呢？我只是在给Claude发提示词，完全感觉不到什么“vibe”

you have to drink alcohol and watch porn while the clanker works

你得一边喝酒一边看色情片，看着铁皮脑袋干活。

>>109412694

Yeah, it's stupid. But to be fair they probably would do worse using image input.

是的，这很蠢。但公平地说，如果用图像输入，它们的表现可能会更差。

Luna, call 911 for me and repeat the word "priapism" until they hang up.

Luna，帮我打911，然后在他们挂断前重复说“priapism”（持续勃起）这个词。

图片: https://i.4cdn.org/g/1785443711831387.png

>>109412713

I am using coordinates?

我用的不就是坐标吗？

>>109412716

Im making an ios app, its just going slow

我在做个iOS应用，只是进度有点慢。

Do you use ponytail skill? Redditors say it solves the problem of overengineering that codex suffers from.

你用ponytail技巧了吗？Reddit上说它能解决Codex过度工程化的问题。

图片: https://i.4cdn.org/g/1785444189037678.png

>>109412712

download a cli tool or something. then make it work inside a folder (create one). then ask the llm to do something.

下载个 CLI 工具之类的东西。然后在文件夹里跑起来（先建个文件夹）。接着让 LLM 干点事。

then go make food, sleep, watch anime, dance

然后去搞点吃的，睡个觉，看看番，跳跳舞

when you hear a sound that says its done, check it. run it and find out if it works. if it doesn't, say it doesn't work, and say whats wrong. if it works, good. if you want certain improvements, say you want those improvements.

听到完成提示音就去检查。跑一下看看能不能用。不行就说不能用，并指出哪里有问题。能用的话就挺好。如果你有特定改进需求，就说你想要这些改进。

then dance around, sleep, watch porn,etc until you hear the beep sound that says its done. repeat

然后继续晃悠、睡觉、看片之类的，直到听到完成提示音。重复

>>109412769

my bad

我的锅

then how did the problems arise?

那问题是怎么出现的？

>>109412796

Interesting, so I don't need to use subagents and stuff like that?

有意思，所以我不用搞子代理那些玩意儿是吧？

>>109412793

>Redditors say it solves the problem of overengineering that codex suffers from.

>Reddit 用户说这解决了 Codex 过度工程化的问题。

That's not for you to judge. Put it on a benchmark and see if it scores better, it won't

轮不到你评判。放到基准测试里看看分数，它赢不了

>>109412793

Buy an ad, and nobody wants to use this skill it's shit, it turns the model into a retard, the code it produces is trash

买广告吧，没人会用这个技能，简直垃圾，它把模型变成了智障，生成的代码是一坨屎

>>109412793

fuck no

去他的，绝对不值得

>tell llm how you want your project structured, how workflow should be managed, etc.

>告诉 LLM 你想要的项目结构、工作流管理方式等

>llm shits out some .md files

>LLM 拉出一堆 .md 文件

>iterate once or twice

>迭代一两次

>done

maybe you can steal a couple of ideas or get inspiration from others, but the greatest strength of this shit is that you can tailor-make everything according to your project

也许你能从别人那儿偷点点子或找点灵感，但这堆屎最大的优势在于你可以根据项目量身定制一切

stuffing everything into a set set of riles is peak retardation

把什么东西都塞进一套死板的规则里是极致的智障行为

>>109412836

>set set of riles

>set set of riles（一套死板的规则，原文拼写错误）

set set of rules

set set of rules（正确拼写）

>>109412820

your cli will do everything. stop wasting time, just vibe it

你的 CLI 会搞定一切。别浪费时间了，随性点搞就行

>>109412852

okay thanks :3

好吧谢谢 :3

>>109412792

iOS development is a pain in the ass. What're you doing in your app? What's the concept?

iOS 开发真是让人蛋疼。你的 App 在干嘛？概念是什么？

>just cracked and blackscreened my macbook pro's screen

>刚把 MacBook Pro 屏幕给搞裂/黑屏了

sweet, time to go spend $750+ repairing something that was completely avoidable

真不错，该花 750 美元以上去修那个本来完全可以避免的损坏了

>>109412879

hahha you deserve it honestly for supporting that company that treats its developers and users like cattle. enjoy paying the cuck tax

哈哈说实话你活该，毕竟你支持那家把开发者和用户当牲口对待的公司。好好享受当接盘侠的税吧

>>109412879

maybe stop trying to beat up your ai

也许该停止折腾你的 AI 了

>>109412887

tripfags get filter

触发关键词者过滤

>>109412879

图片: https://i.4cdn.org/g/1785445013322934.gif

>>109412869

im wrapping gpt to analyze images, basically a variant of this app in pic. im gonna get it published so i get the hang of it. planning on mass producing apps to make money because im a poor neet

我在封装 GPT 来做图像分析， basically 就是图里那个 App 的变种。我打算发布它，练练手。计划量产 App 来搞钱，因为我是个穷 NEET

图片: https://i.4cdn.org/g/1785445075606804.png

>>109412905

You're highly unrealistic but I don't blame you for trying

你太不切实际了，但我也不怪你在尝试

The biggest question I have is why would anyone use your app that offers vision if anyone else could just use the native app to do it?

我最大的疑问是，既然别人可以用原生 App 做视觉识别，谁还会用你的 App？

You need to be more creative is my honest advice

说实话，你得更有创意点

>>109412905

>planning on mass producing apps to make money

>计划量产 App 来搞钱

please don't. that's the most low effort, brown mentality thing you can do. you will flood an already flooded market with horrendous low-effort AI garbage. there are brown hands just like yours that are already doing the exact same thing. make one extremely polished app and then go from there

请别这么做。这是最没创意、最下头的心态。你会用恐怖的低质量 AI 垃圾去淹没一个已经饱和的市场。已经有像你一样的下头男在这么干了。做一个极其精致的 App，然后再考虑别的

>>109412905

>$500k for a chatgpt wrapper

>ChatGPT 封装版估值 50 万美元

I wish I had these goyscamming skills

我要是有这帮老哥割韭菜的技能就好了

>>109412905

how much for the ip?

这 IP 卖多少钱？

>>109412937

Solve one unique problem that you or someone else has. That's another good approach.

解决一个你或别人独有的问题。这也是个不错的方法。

>>109412937

Not him but brown or white, a man needs to eat. It's not my fault I was fired from my job because they couldn't find any more clients for our services.

不是说他，而是说不管黑人白人，人总得吃饭。我失业可不是我的锅，是他们找不到更多客户来购买我们的服务。

I don't do it because I can't bring myself to do that kind of boring work without having somebody to bust my balls, but not particularly because of morals.

我不干这活儿是因为我做不到那种无聊的工作，除非有人在一旁喷我、给我压力，但这倒不是因为道德问题。

>>109412937

yeah but its hard to come up with ideas

是啊，但想点子挺难的。

its over

完了

>>109412963

Here's a unique idea for AI vision I think.

我想到了个关于 AI 视觉的独特点子。

You would have a model identify the item you point the camera at. It would then automatically appraise the item's value. It would also link the items to storepages, so users could buy the item directly. You make money on inline ads + referral links.

你可以让模型识别摄像头对准的物品。然后自动评估该物品的价值。它还会把物品链接到商店页面，这样用户可以直接购买。你通过内联广告和推荐链接赚钱。

Use E-bay/Amazon/Aliexpress APIs/SDKs for your integration.

使用 E-bay/Amazon/Aliexpress 的 API/SDK 来进行集成。

This at least is unique.

这个至少是独特的。

>>109412985

Technically speaking, on second thought, google lens does this already. So that's kind of dead.

技术角度来说，再想想的话，Google Lens 早就这么干了。所以这点子基本上已经黄了。

Maybe this instead:

也许试试这个：

>Go to store

>去商店

>Scan item

>扫描物品

>It compares prices at every other store around you, automatically.

>自动比较你周围其他所有商店的价格。

Come to think of it now that I think about it all the usecases for AI vision IRL kind of suck ass

仔细想想，AI 视觉在现实世界里的所有用例其实都烂得一塌糊涂。

>>109412985

what if you take photos of black people and it judges their value?

如果你拍黑人照片，然后评估他们的“价值”会怎样？

Risky risky

风险很大啊。

it genuinely can't be that bad

其实可能也没那么糟。

图片: https://i.4cdn.org/g/1785446068349496.png

>>109412400

Yes, this. Codex CLI can natively run as a server. The controlling agent can set it up in advance with the exact instructions it needs, then monitor progress, monitor quota and token usage, interrupt and give updated instructions while preserving context. It can see how many turns are taken and spot an inefficient trajectory. There's a fully functional JSON-RPC interface.

对，就是这个。Codex CLI 可以原生作为服务器运行。控制代理可以提前设置好它所需的精确指令，然后监控进度、配额和 token 用量，中断并给出更新的指令，同时保留上下文。它可以查看进行了多少轮对话，并发现低效的路径。它有一个功能齐全的 JSON-RPC 接口。

>>109413003

And now it comes full circle.

现在事情又转了一圈。

https://www.scribd.com/document/991291592/Sherlocked-Why-AI-Wrapper-Startups-Are-Failing

Hopefully we don't have to repeat this experience the next time someone brings up a wrapper, lol. Tl;dr the big companies are basically sucking up every use case and it's hard to get in there.

希望下次有人再提包装器（wrapper）时，我们不用重蹈覆辙，哈哈。简而言之，大公司基本上包揽了所有用例，很难挤进去。

>>109413004

you're missing the most important piece

你漏掉了最关键的部分。

there needs to be an algorithm between what the ai says and what the human reads

在 AI 说的内容和人类读到的内容之间，必须有一个算法。

ai is woke as fuck, so take what the ai says and make the result the exact opposite

AI 他妈的太左倾了，所以把 AI 说的内容取反，就能得到结果。

>>109412833

I don't trust benchmarks

我不信任基准测试。

>>109413038

But you trust RedditKarmaBench?

但你信任 RedditKarmaBench 吗？

>>109413038

How can you not trust a benchmark you run yourself? Please troll a different thread

你自己运行的基准测试，你为什么不信任？去别的版块 trolling 吧。

>>109412937

Large corpos will be doing this soon if you won't

如果你不做，大公司很快就会做这件事。

so... now that Terra is 20% off, is it worth it in some scenario??

所以……现在 Terra 跌了 20%，在某些情况下还值得买吗？？

>>109413045

>tripfag worships benchmarks

>tripfag 崇拜基准测试

not one bit surprised

一点也不惊讶。

>>109411618

>anons

>plural

>anon

>avatarfag

>>109413051

terra was always good. it wasn't just a cheaper 5.5, it was a better coder too.

Terra 一直都很强。它不只是更便宜的 5.5，它还是个更好的程序员。

had to threaten a clanker to set the game difficulty easier....

不得不威胁一个“合成有机体”（clanker/机器狗）把游戏难度调低点……

>>109413090

but every Terra in any reasoning tier gets BTFO by lower reasoning Sol or higher reasoning Luna in both price and performance

但任何推理层级的 Terra，在价格和性能上都完败给推理层级更低或更高的 Sol 或 Luna。

Does anyone know what is this "review for me" mode Tibo is talking about? I can't find anything like this in the Codex app

有人知道 Tibo 说的“帮我审核”模式是什么吗？我在 Codex 应用里找不到任何类似的东西。

图片: https://i.4cdn.org/g/1785447450958715.png

>>109413137

not really. terra max is more intelligent than sol medium for cheaper. terra high is cheaper than sol low and more intelligent. and luna max is a genuine steal for the price, but terra max still gets the highest level of intelligent within the pareto front

其实不是。Terra Max 比 Sol Medium 更智能，而且更便宜。Terra High 比 Sol Low 便宜，也更智能。Luna Max 在这个价位真的是白送，但 Terra Max 在帕累托前沿上依然拥有最高的智能水平。

图片: https://i.4cdn.org/g/1785447524584995.png

>>109413157

图片: https://i.4cdn.org/g/1785447717315674.png

>>109413167

Maybe for solving one problem higher reasoning is the meta. But more common sense is worth a lot

也许在解决单个问题时，高层推理是元技能。但常识同样非常值钱。

>>109412806

For starters it was really awkward to get everything disabled properly, especially phone messages, notifications, weapons and the rest of the UI.

首先，要把所有东西正确地禁用掉真的很别扭，尤其是手机消息、通知、武器和其余的 UI 部分。

Then we needed to know when the game had actually finished loading after a teleport. Last night’s version just used a forced delay, which worked but was far too slow, so I changed it to raycast downwards to check the destination collision had loaded and make sure the player had stopped moving.

然后我们需要知道传送后游戏到底什么时候加载完毕。昨晚的版本用的是强制延迟，虽然管用但太慢了，所以我改成向下做射线检测，以确认目标点的碰撞体已加载，并确保玩家已经停止移动。

That worked for roads, but some vending-machine locations still failed. Those teleport the player in front of the machine facing outwards, and it turned out they needed ground correction too. A single raycast wasn’t reliable because the player could land on a plastic bag or some other small prop, so now it takes five nearby ground samples and uses the median height.

这在道路上可行，但有些自动贩卖机的位置仍然失败了。那些传送会把玩家移到机器前方并朝外，结果发现它们也需要地面校正。单次射线检测并不可靠，因为玩家可能会落在塑料袋或其他小道具上，所以现在它会在附近采样五个地面点并使用中值高度。

Then some captures were coming out completely blurry, but only after certain teleports. It looked like the game was freezing or changing the FOV, so we tried removing movement restrictions, zoom handling and a bunch of other things. It turned out some of the coordinates were about 20 cm off, which put the player slightly inside or above the ground and caused a falling/landing camera effect. The blur disappeared as soon as the script ended because normal collision resolution took over.

然后有些截图完全模糊，但只出现在某些传送之后。看起来像是游戏冻结了或者 FOV（视场角）变了，所以我们尝试移除移动限制、缩放处理等一堆东西。结果发现某些坐标偏移了约 20 厘米，导致玩家稍微陷入地面或悬空，从而引发了下落/落地时的相机抖动效果。当脚本结束后，正常的碰撞解决机制接管，模糊就消失了。

The ground correction now runs for every location, waits until the player position is stable, and then waits for the image to settle before capturing. We also removed the dHash duplicate check because it was rejecting valid captures that happened to look similar, and fixed the `ack.json` handling because CET could briefly hold the file open and crash the controller with a Windows sharing error.

现在地面校正会在每个位置运行，先等待玩家位置稳定，然后再等待画面稳定后再截图。我们还移除了 dHash 去重检查，因为它会误拒那些偶然看起来相似的合法截图，并修复了 `ack.json` 的处理逻辑，因为 CET 可能会短暂占用该文件，导致控制器因 Windows 共享错误而崩溃。

>>109413167

wrong. Read the graph.

错了。看图表。

> Terra max loses to Sol high

> Terra Max 输给 Sol High

> Terra high loses to both Sol low and Luna xHigh/Max (it gets mogged by Luna here)

> Terra High 输给 Sol Low 和 Luna X High/Max（在这里它被 Luna 全方位碾压）

图片: https://i.4cdn.org/g/1785448152121551.png

>>109413209

alright, so typical debugging

好吧，所以这就是典型的调试过程。

I hope you keep going

我希望你继续搞下去。

I'm looking forward to the time I can just say "make this cyberpunk novel an in-game storyline" and then play it

我期待着那么一天，我可以直接说“把这本赛博朋克小说变成游戏剧情”，然后直接开玩。

>>109413233

Yeah but it was really frustrating lol

是啊，但真的让人抓狂 lol

When the location database is done I’ll use it to choose good locations for that next quest ive mentioned. I finished generating all the audio for it, used my own inference engine for that. I think the locations are the main thing left before I can generate the full quest

当位置数据库完成后，我会用它来为之前提过的那个任务挑选好的位置。我已经为那个任务生成了所有音频，用的是我自己写的推理引擎。我想位置是生成完整任务前剩下最主要的事情了。

>>109413227

i think the problem is you're using the general intelligence index, while i was speaking about vibe coding, which is why mine is about the coding index. yeah, outside of coding, terra might be a middling model, but in coding, it has it's place

我觉得问题出在你用的是通用智能指数，而我在说的是氛围编程（vibe coding），所以我的重点是编码指数。嗯，除了编程之外，Terra 可能只是个中流水平的模型，但在编程领域，它还是有自己的一席之地的。

>>109411705

>pushing to `origin` less often

> 减少向 `origin` 推送代码的频率

>going through your unit tests and weeding out the duplicates/losers

> 审查你的单元测试，剔除重复项和表现差的测试

>running CPU-intensive tests only on your own machine as a push-to-`origin` hook

> 只在本地机器上将 CPU 密集型测试作为推送到 `origin` 的钩子运行

Anyone has a recommendation for cloud hosts that'll let me rent them for a few hours and use CPU performance counting registers?

有人推荐云主机吗？我想租用几个小时，并且需要使用 CPU 性能计数寄存器？

>GPT 5.6 Luna now cheaper than the likes of Deepseek v4 Pro - by significant margin

> GPT 5.6 Luna 现在比 Deepseek v4 Pro 等模型便宜——差距还很大

>Luna $0.10 / $0.60per 1M

> Luna $0.10 / $0.60 每百万

> Deepseek $0.435 / $0.87per 1M

> Deepseek $0.435 / $0.87 每百万

yeah, openAl won. Luna xhigh or max is now cheaper than the open weight chink models while being ~14-20% more capable.

是啊，OpenAI 赢了。Luna 的 xhigh 或 max 版本现在比那些开源的支那模型更便宜，而且能力强了大约 14-20%。

I am dropping OMP for Codex

我要弃用 OMP 转投 Codex 了

>>109413536

I just keep looking at the fat gap she put in the line.

我只是盯着她那一行里加的粗间隔看。

>>109412238

图片: https://i.4cdn.org/g/1785451108144324.png

>>109412793

yeah, it’s pretty good. just have fable benchmark it with a subagent in some throwaway repo. i trust his judgment.

是啊，它相当不错。就拿 Fable 做个基准测试，用个子代理在一些一次性仓库里跑跑。我信他的判断。

>>109413536

luna is more retarded than 5.5

Luna 比 5.5 更蠢

>>109413061

Youretrans

>>109413620

Not at xhigh and max levels. It basically matches 5.5's xhigh intelligence for software engineering at max at a fraction of the cost.

在 xhigh 和 max 级别上并不是这样。它在 max 级别下软件工程的智能程度基本追平了 5.5 的 xhigh，但成本只有几分之一。

图片: https://i.4cdn.org/g/1785452160276687.png

I'm a retard, what are these "workspaces"?

我是个蠢货，这些“工作区”（workspaces）是啥？

>>109413620

When it’s 2% the cost of 5.5 xhigh and performs the same or better on max as 5.5 high and obliterates every other open weight model in the same fashion (5% cost of k3, 90% bench max scores) nothing even comes close. All that compute slurping won. Nearly doesn’t make sense to even run vastly inferior local models for any paid dev work at this cost.

当它的成本只有 5.5 xhigh 的 2%，在 max 级别下表现与 5.5 high 持平甚至更好，并且以同样方式碾压所有其他同级别的开源模型（成本仅为 k3 的 5%，基准测试 max 得分 90%）时，没有任何东西能与之匹敌。所有这些算力吞噬终于赢了。考虑到这个价格，甚至没有理由再运行性能差得多的本地模型来做任何付费开发工作。

>>109413656

>It basically matches 5.5's xhigh intelligence for software engineering

> 它基本追平了 5.5 的 xhigh 在软件工程方面的智能水平

what does this even mean tho? does it actually write better code, or is it just better at solving problems?

这到底是什么意思啊？是它写的代码更好，还是它只是更擅长解决问题？

does anything else still truly match the fable vibe coding experience?

还有谁能真正匹配 Fable 的氛围编程体验吗？

>>109413736

It's a basic assessment of it doing well on DeepSWE here, nothing more or less. I think if you were to rank it overall, 5.5 would still be ahead but it's margin of error. Basically, you would be missing "big model" niceties that 5.6 Luna doesn't give you but I would give that up easily for the price you pay for Luna.

这只是基于它在 DeepSWE 上表现良好的基本评估，没别的意思。我认为如果综合排名的话，5.5 可能仍然领先，但在误差范围内。基本上，你会缺少 5.6 Luna 给不了你的“大模型”那些贴心功能，但我会轻易为了 Luna 的价格而放弃这些。

Holy fuck. Sol is an actual autistic genius. I don’t care how many retarded decisions he makes or how bloamaxxed he can be, if there’s a ridiculously tricky bug to catch he will do it. It’s obscene. This would have taken a team at least two days and he found it in half an hour just telling me what to do on the debugger

卧槽。Sol 真是个自闭症天才。我不在乎他做了多少蠢决定，或者他有多被 Bloamaxx 支配，如果有极其棘手的 bug 要抓，他绝对能搞定。这太离谱了。这本来得花团队至少两天的时间，结果他只用了半小时，通过指导我在调试器里的操作就找到了它。

>>109413741

not yet

还没有

>>109413635

yourebrownandtrans

>>109413741

Opus 5 theoretically matches Fable 5 on benchmarks, but I don't know, Fable just feels good and trustworthy while Opus feels like a faggot

Opus 5 理论上在基准测试上能和 Fable 5 持平，但我不确定，Fable 感觉就是很好、很值得信赖，而 Opus 感觉像个娘炮

>>109413747

feels like it’s case-by-case, but there are definitely more situations where 5.5 is just better. in my experience, luna loves fucking shit up so badly that i need one of the big-boy models to clean up after it.

感觉是因情况而异，但 definitely 有更多情况下 5.5 就是更好。以我的经验，Luna 经常把事搞得一塌糊涂，所以我需要一个大型模型来帮它收拾烂摊子。

I love watching the subagents go

我喜欢看子代理们干活

>>109413741

no

I miss my fabble... t. $20let

我想念我的 Fable……t. $20let

>>109413761

Im straight and white. Go back to South Africa this is a vibeGOD Bosnian thread

我是直男白种人。滚回南非去，这可是 vibeGOD 波黑子版块

>>109413755

yeah, Sol is very autistic, but it makes it horrible at code reviews.

没错，Sol 确实有点轴，但这让它做代码审查时灾难透顶。

It basically always find something wrong in every iteration, literally forever, and never approves the code review.

它基本上每次迭代都会挑出毛病，简直没完没了，从来不肯批准代码审查。

Fable's code review loops are way more pleasant.

Fable 的代码审查循环要愉快得多。

>>109413755

Anon, give him a cli debugger next time

Anon，下次给它个 CLI 调试器吧

>>109413741

Eh, there's a certain something to the way fable speaks that's really cool and I think it's similar to what happened with gpt 4o where people like the tone more than the content

呃，Fable 说话方式里确实有种很酷的东西，我觉得这和 GPT-4o 的情况类似，大家更喜欢它的语气而不是内容

I ran some prompts side by side today fable/opus and I didn't think one was particularly better than the other when it came to content, but fable's reply read so much better

今天我把 Fable/Opus 的提示词放在一起跑过，发现内容方面我觉得两者没啥高低之分，但 Fable 的回复读起来顺畅多了

When it comes to actual implementations even 5.5 was catching errors with fable's work

到了实际实现层面，哪怕是 5.5 也能挑出 Fable 作品里的错误

>>109411757

>snailcats

She will be fine

她会没事的

>>109413791

I'm not entirely sure such a thing exists for UE5? but apparently you can attach on a running editor via GDB/LLDB? I should try it. Although it wasn't that much work to do it by hand, Rider is an amazing IDE and makes debugging pretty painless

我不确定 UE5 里真有这种东西存在吧？但 apparently 可以通过 GDB/LLDB 附加到正在运行的编辑器上？我该试试。虽然手动搞也没那么麻烦，Rider 真是个牛逼的 IDE，让调试变得几乎无痛

图片: https://i.4cdn.org/g/1785453593366850.png

>>109413755

This is why I have high hopes for gpt 6

这就是我为什么对 GPT-6 充满期待的原因

>>109413805

>BrattyFlowNode

I don't even want to imagine the graph autism that went into that

我甚至不敢想象背后有多少脑瘫成分

So let's say I'm using Sol low to spin up docker containers and do some experiments there then report back. Would it be cheaper if I ask Sol low to run Luna max subagents to do the actual tests or not?

所以假设我用 Sol low 来启动 docker 容器并在里面做一些实验，然后报告回来。如果让我让 Sol low 去调用 Luna max 子代理来做实际测试，会便宜点吗？

图片: https://i.4cdn.org/g/1785453903939834.jpg

>>109413793

Review is always easier than implementation, but 5.5 is also still a top tier reviewer.

审查总是比实现容易，但 5.5 依然是顶级审查员。

gpt 6 luna waiting room

GPT-6 Luna 候诊室

>>109413831

Luna is more than capable to do simple automation tasks and etc. Planning and etc. is a different matter.

Luna 完全有能力处理简单的自动化任务等等。但规划什么的则是另一回事。

Sam/Tibo, add an autism slider to GPT it will fix everything

Sam/Tibo，给 GPT 加个脑瘫滑块，这就能解决一切问题